Total variation noise reduction algorithm in computed tomography image with custom-built phantom using 3D-printer

Abstract

Noise in computed tomography (CT) is unavoidable because of various factors such as patient-source-related errors, hardware error, and electrical interference, which lead to unwanted diagnosis errors. Therefore, to solve this problem, we model a noise reduction algorithm based on total variation (TV) and applied it to images acquired using simulation study and 3D printing technology. Moreover, the conventional noise reduction algorithms are applied to the same image for comparative evaluation. For quantitative evaluation of the algorithms, we use the parameters of the coefficient of variation, signal-to-noise ratio, contrast-to-noise ratio, and normalized noise power spectrum. Our results indicate that the proposed TV noise reduction algorithm affords greater improvement in all the evaluation parameters considered in the simulation and 3D-printed phantom study, over the conventional noise reduction algorithms. In conclusion, we believe that our approach can significantly contribute to CT study and application.